# Virtuals says Robinhood opened agentic trading to all eligible users > Source: > Published: 2026-09-07 00:52:12+00:00 # Virtuals says Robinhood opened agentic trading to all eligible users **Robinhood launched Agentic Trading in May; nearly 100,000 accounts had opened by late June, while Vlad Tenev said setup still favors technical users.** By [RuntimeWire Staff](/author/runtimewire-staff) ยท Published Primary source: [Virtuals on X via Aligned News](https://x.com/virtuals_io/status/2096605353656168513) ## Why it matters Robinhood is turning MCP into a retail brokerage interface, letting outside AI providers generate real orders. The upside is a new distribution channel; the unresolved risk is how responsibility for an erroneous trade works in practice among Robinhood, the agent provider and the customer. Virtuals said Robinhood, led by co-founder and CEO Vlad Tenev, has completed the rollout of [Agentic Trading](https://robinhood.com/us/en/agentic-trading/?ref=runtimewire) to all eligible users. Robinhood's own materials verify that the brokerage launched the product on May 27th and subsequently expanded its capabilities, though they do not establish when access reached every eligible customer. The rollout advances a bet from Robinhood co-founder and CEO [Vlad Tenev (@vladtenev)](https://x.com/vladtenev?ref=runtimewire): the next generation of brokerage customers will include software acting under instructions from humans. Robinhood built a dedicated account for those agents rather than adding another market-analysis chatbot to its app. An outside agent can inspect a customer's portfolio, research securities, build or rebalance positions and submit orders through Robinhood's Model Context Protocol, or MCP, server. That gives Tenev's AI thesis a distribution channel through the brokerage he started with [Baiju Bhatt (@BaijuBhatt)](https://x.com/BaijuBhatt?ref=runtimewire) in 2013. The two met at Stanford, studied mathematics and physics, and co-founded Celeris, a high-frequency-trading software company, before turning their attention to retail investors. Bhatt stepped away from Robinhood's operating team in March 2024, and [Robinhood's current board biography](https://investors.robinhood.com/governance/board-of-directors?ref=runtimewire) lists him as a director. Tenev, who studied mathematics at Stanford and UCLA, also co-founded AI reasoning developer Harmonic and serves as its executive chairman. Tenev has described Robinhood's original mission as widening access to investing for people without substantial wealth. When Robinhood [introduced Agentic Trading on May 27th](https://robinhood.com/us/en/newsroom/robinhood-is-now-open-to-agents/?ref=runtimewire), he extended that language directly to AI agents. The product makes the brokerage account programmable, shifting part of Robinhood's interface from screens designed for people to tools that software can call. ### A brokerage account built for software Customers connect a third-party agent to a dedicated Robinhood Agentic Account through the brokerage's MCP server. Robinhood's [setup documentation](https://robinhood.com/us/en/support/articles/agentic-trading-overview/?ref=runtimewire) lists Claude Code, Claude Desktop, ChatGPT, Codex, Cursor and Grok among the compatible platforms, while allowing other MCP-compatible services to connect. The agent receives read access to account numbers, balances, positions, transaction history, watchlists and scans across the customer's Robinhood accounts. Its trading authority is limited to the separate Agentic Account. Customers decide how much money to transfer into that account, receive trade notifications and can disconnect the agent. Robinhood's materials describe support for equities, options and crypto. Crypto availability remains subject to account and state restrictions, and agents cannot transfer, stake or lend crypto. Robinhood says customers can ask an agent to assemble a portfolio, automate a trading rule or rebalance existing positions. If a customer instructs the agent to act without asking permission, the agent can place trades without transaction-by-transaction confirmation. That last capability is the product. It is also where Robinhood's disclosures become more direct than its launch language. Robinhood warns that an agent can misunderstand instructions, use incomplete or outdated information and behave unexpectedly. Customers assume responsibility for the resulting trades and for data shared with the outside AI provider. Robinhood says it does not control, supervise, monitor or audit those agents. The separate account limits the amount of capital exposed to an agent, but it does not remove execution risk inside that account. Robinhood also warns that AI strategies can move quickly and may be difficult to stop in real time. The design puts a budget around automation while leaving the customer responsible for deciding whether the software can be trusted with it. ### Adoption arrived before simplicity Robinhood reported meaningful early uptake. In its [second-quarter results](https://investors.robinhood.com/news-releases/news-release-details/robinhood-reports-second-quarter-2026-results?ref=runtimewire), Robinhood said nearly 100,000 customers had opened Agentic Trading accounts holding more than $100 million in assets under custody. Those figures imply assets on the rough order of $1,000 per opened account, though Robinhood's rounded thresholds prevent an exact calculation. The account count also measures openings rather than active agents, retained users or completed trades. Robinhood did not attach Agentic Trading revenue, trading volume or usage frequency to the disclosure, leaving the commercial value of those accounts separate from the initial adoption number. Tenev acknowledged the onboarding problem during Robinhood's [second-quarter earnings call](https://investors.robinhood.com/static-files/f44ea493-9c52-4744-9671-7720e1ac7b2f?ref=runtimewire). He told analysts that connecting Robinhood with tools such as Codex or Claude Code still required considerable technical sophistication. Some models were also reluctant to execute trading instructions, he said, forcing users to work around an assistant's built-in caution. That friction explains why a completed eligibility rollout would mark distribution progress rather than a finished consumer product. Robinhood can open the feature to its customer base while the practical user group remains concentrated among people comfortable configuring MCP connections, desktop developer settings and command-line tools. Robinhood has an incentive to close that gap. The brokerage [reported 27 million funded customers at the end of 2025](https://investors.robinhood.com/news-releases/news-release-details/robinhood-reports-second-quarter-2026-results?ref=runtimewire). Agentic Trading had therefore reached only a small fraction of Robinhood's funded customers despite a fast start for a product introduced two months earlier. ### Tenev is opening Robinhood to outside agents Robinhood's strategic choice is the bring-your-own-agent model. [Webull's agentic tools](https://www.webull.com/agentic?ref=runtimewire) also connect MCP-compatible agents to brokerage accounts, while [Public's in-app agents](https://help.public.com/en/articles/15435640-agents-the-basics?ref=runtimewire) can research markets and execute user-defined plans. [Alpaca's trading infrastructure](https://alpaca.markets/blog/alpaca-introduces-cli-for-trading-api/?ref=runtimewire) gives developers API and command-line tools for connecting AI agents to trading, market-data and account-management functions. Robinhood is betting that customers will want to choose their preferred AI provider while keeping assets and execution inside a familiar brokerage account. That choice gives Robinhood broad compatibility and avoids requiring one in-house model to win every customer. It also divides the experience among Robinhood, the agent provider and the customer's instructions. Robinhood's disclosures say customers are responsible for trades placed by their agents and for losses from agent-generated decisions. The operational allocation of responsibility among Robinhood, the outside agent provider and the customer remains unclear when an erroneous trade occurs, even though Robinhood supplies the account and executes the order. The rollout puts Robinhood in position to capture orders generated by whichever AI interfaces customers adopt, while the dedicated account gives cautious users a way to test automation with a defined pool of capital. Broad availability will make the next set of numbers more useful than the first. Account openings showed curiosity. Assets, recurring use and trading activity will show whether customers are willing to let outside software keep its hand on the order button.